Per-target affinity scoring, maturation scans, calibration status, and wet-lab readiness in one workspace.

Affinity is the bridge between structure-first ranking and wet-lab measured KD. The page makes model readiness, target coverage, and data gaps visible.

Candidate set

Sequences or existing candidates to score.

Target

Target name must match available calibration/selection data for the strongest output.

Parent sequence

For maturation scans, the parent binder sequence and mutation scope.

Affinity predictions

Predicted KD or ranking values when the target/model path supports it.

Maturation suggestions

Mutants scanned around a parent with predicted deltas and limitations.

Readiness status

Which scorers are ready, need weights, need config, or remain stubs.

1. 1. Open the status panel first

   Open the status panel first. Confirm which scorers and targets are ready.

2. 2. Use candidate scoring when you have a cohort to rank

   Use candidate scoring when you have a cohort to rank.

3. 3. Use maturation scan when you have a parent binder and want local mutations

   Use maturation scan when you have a parent binder and want local mutations.

4. 4. Treat limitations as part of the result

   Treat limitations as part of the result. Missing calibration is a decision blocker, not fine print.

• GET /api/v1/affinity/status reports scorer readiness and recent runs.
• Candidate scoring and maturation panels use lib/api job helpers and poll until terminal.
• Modal/internal GPU paths exist for configured scorers; readiness varies by target.

• A model marked needs_weights or needs_config is not production evidence.
• Predicted KD should be calibrated against wet-lab data before being used as a hard gate.
• Affinity and specificity are different questions; use Interface and counter-screens together.